Clean, copy-paste tutorials for building AI agents: first-class code blocks, real eval results, and no fluff. Follow a path or jump straight to the step you need.
$ npm i && copy-paste the next step
Each path is an ordered run of copy-paste tutorials. Start at step one or drop in wherever you are.
Build a working agent loop with no framework: tool calls, memory, and control flow you fully own.
Drop an agent into an app you already ship, without rewriting your codebase or your data model.
Coordinate specialist agents: routing, hand-offs, and shared state across a multi-agent system.
Measure agents like software: metrics, regression suites, and traces that catch silent failures.
Single, self-contained walkthroughs you can finish in one sitting.
allowedTools auto-approves, tools restricts. That half is documented. What is not: an AgentDefinition has no allowedTools field, and the SDK forwards one verbatim to the CLI anyway, unvalidated.
A grep bisect tells you when today's code landed, not when the behaviour changed. Here is the argv-capture harness that dates it properly, and the worked example where grepping made a maintainer look wrong and running the versions proved him right.
The skills option validates every name before the session starts, and its wildcard guard tests for exactly two endings. Eight of eleven wildcard shapes get through, become a Skill(name) rule that names no skill, and say nothing. Measured first-party on 0.3.272, with the bisect.
Since @anthropic-ai/claude-agent-sdk 0.3.257 a malformed hook entry is deleted from settings.json and the rest of the file loads normally, so the hook just never runs. PreToolUse and PermissionRequest are the two exceptions, where the whole file is voided. Bisected against 0.3.252, with a runnable detector.
npm error notarget No matching version found for @anthropic-ai/sdk means you asked for a major that does not exist. Anthropic ships two SDKs on unrelated version lines: Python is on 1.5.0, npm has never left 0.x. Read from both registries, with the caret trap that follows.
TypeError: Invalid http_client argument means the anthropic Python SDK moved to httpx2 in 1.0.0. The rename is one import. The parts nobody mentions are that the guard printing that message only exists from 1.4.0, and that it cannot see your respx or OpenTelemetry layer at all (2026).
Tell us your model provider and runtime, and we'll surface the tutorials that match.
The latest tutorials, fresh off the press.
allowedTools auto-approves, tools restricts. That half is documented. What is not: an AgentDefinition has no allowedTools field, and the SDK forwards one verbatim to the CLI anyway, unvalidated.
A grep bisect tells you when today's code landed, not when the behaviour changed. Here is the argv-capture harness that dates it properly, and the worked example where grepping made a maintainer look wrong and running the versions proved him right.
The skills option validates every name before the session starts, and its wildcard guard tests for exactly two endings. Eight of eleven wildcard shapes get through, become a Skill(name) rule that names no skill, and say nothing. Measured first-party on 0.3.272, with the bisect.
Found a bug in a tutorial, want a topic covered, or shipping something with these patterns? Send the details. Code snippets and repro steps welcome.
